Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
EC 613:Advanced Topics in Financial Econometrics
FEC 514:Applications in Financial Modeling
Harald Schmidbauer
c© Harald Schmidbauer & Angi Rosch, 2012
About These Slides
• The present slides are not self-contained; they need to be explained anddiscussed. This will be done in the lectures.
• Even though being a “work in progress” and subject to revision, the slidesconstitute copyrighted material.If you want to reproduce or copy anything from the slides, please ask:
Harald Schmidbauer harald at hs-stat dot comAngi Rosch angi at angi-stat dot com
• The slides were produced using LATEX and R (the R project; website: www.R-project.org) on a GNU/Linux system.
• R files used for this course are available upon request.
c© Harald Schmidbauer & Angi Rosch, 2012 About these slides 2/32
Some Projects
From Our Recent Research
c© Harald Schmidbauer & Angi Rosch, 2012 Recent Research Projects 3/32
Recent (and ongoing) research projects.
The following slides give an outline of four projects.
• Project 1: Green segmentation: a cross-national study
(with Barıs Yılmazsoy & Angi Rosch)
• Project 2: Algorithmic trading
(with Angi Rosch, Tolga Sezer & Vehbi Sinan Tunalıoglu)
• Project 3: Festivals and Gold Prices
(with Angi Rosch)
• Project 4: Population Dynamics With Leslie-Type Models
(with Angi Rosch & Narod Erkol)
• Project 5: OPEC Announcements and Oil Price Volatility
(with Angi Rosch)
c© Harald Schmidbauer & Angi Rosch, 2012 Recent Research Projects 4/32
Project 1:
Green Segmentation: A Cross-National Study
Some aspects.
• world facing environmental challenges
• business consequences
• shift in consumer attitude and preferences
• understanding the “green” consumer is important
• key concepts: attitudes, behavioural intentions
• theory of Reasoned Action:
attitudes ⇒ behavioural intentions ⇒ actual behaviour
• relationship between green attitudes and green behaviour:
no agreement in literature
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 1 5/32
Project 1:
Green Segmentation: A Cross-National Study
The questionnaire: Attitude items.
strongly agree / agree / indifferent / don’t agree / don’t agree at all
• Tenor, basic attitudes: A1, A4, A9
too much trouble; not too late to save the environment; natural resources are
scarce
• Emotional concern: A2, A5, A8
responsible for global warming; frustrated with industies polluting; frightened
with chemicals in food
• Scepticism: A3, A6, A7
job loss; green trend is marketing gimmick; benefits of consumer products
more important
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 1 6/32
Project 1:
Green Segmentation: A Cross-National Study
The questionnaire: Behaviour items.
always / mostly / sometimes / rarely / never
• Daily behaviour: B1, B3, B8, B9
using public transport; re-usable bags; energy-saving light bulbs; recycling
• Consumption: B5, B6
products with less packaging; buying more expensive “greener alternative”
• Environmental activism: B2, B4, B7
advising others; participation in meetings; reading about environmental issues
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 1 7/32
Project 1:
Green Segmentation: A Cross-National Study
Data & Analysis.
• Data (convenience sample) from China (395), Germany (360),
Turkey (660)
• students enrolled in business-related programmes in
private/public universities
• “Split the difference”-method used to eliminate gender effect
• For each country: four segments (clusters) obtained using “pam”
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 1 8/32
Project 1:
Green Segmentation: A Cross-National Study
Characterizing clusters.
● ●
●
●
●
cndetr
1
1
1
22
2
3
3
34
4
4
attitude
beha
viou
r
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 1 9/32
Project 1:
Green Segmentation: A Cross-National Study
Managerial Implications.
• Focus on segment-specific customer expectations.
• Incentive for shifting customers to “greener segment”?
• “Positive marketing”: parties exchange value such that they are
better off
• Environmental attitude and purchase channel choice???
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 1 10/32
Project 2:
Algorithmic Trading
Algorithmic trading.
• Can we combine simple trading rules to generate profit?
• Example: e-$ trading.
– Trading every 5 minutes.
– Use 2 days’ data to construct a trading rule.
– Use this rule for the next day.
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 2 11/32
Project 2:
Algorithmic Trading
A sequence of trading signals.
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 2 12/32
Project 2:
Algorithmic Trading
How to obtain complex trading rules?
• Genetic algorithm: based on
– cross-over
– mutation
– reproduction
• Example of a program:
(IF TR1 = “buy!”) AND NOT
((TR2 = “sell!”) OR (TR3 = “sell!”)
)THEN “buy!”
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 2 13/32
Project 2:
Algorithmic Trading
Creating a Program.
program 1...
program N
p1...
pM
FC
FC
final program
pool population
No!
Yes!
Yes!
No!
currentfittest
terminalcondition?
fitter?
genetic operations:
• cross-over
• mutation
• reproduction
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 2 14/32
Project 2:
Algorithmic Trading
Keywords.
• data-snooping bias
• robustness
• time series bootstrap
• tick data
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 2 15/32
Project 3: Festivals and Gold Prices
A festival calendar.
2012201120102009200820072006200520042003200220012000199919981997199619951994199319921991
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Aks
haya
Trit
iya
Chi
nese
New
Yea
r
Diw
ali
Dus
sehr
a
Eas
ter
Eid
al−
Adh
a
Ram
adan
Eid
Chr
istm
as
. . . and their impact on daily returns on gold?
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 3 16/32
Project 3: Festivals and Gold Prices
Gold prices.
050
010
0015
0020
00
US
D/o
z.
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
Akshaya TritiyaChinese New YearChristmasDussehraRamadan Eid
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 3 17/32
Project 3: Festivals and Gold Prices
Daily returns on prices.
−8
−6
−4
−2
02
46
8
daily
ret
urn
in p
erce
nt
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
Akshaya TritiyaChinese New YearChristmasDussehraRamadan Eid
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 3 18/32
Project 3: Festivals and Gold Prices
The model: regression and GARCH.
rt = c+ νt√ht = c+ εt (1)
νt = ηt +∑i
bi dit (2)
ht = α0 + α1 ε2t−1 + β ht−1 +
∑i
γi dit (3)
• Summation is over festivals.
• (rt): series of daily returns on gold price
• (dit): (extended) dummy variables for festival i
• (ηt): Gaussian white noise with var(νt) = 1
• bi, γi: parameters quantifying the impact of festival i
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 3 19/32
Project 3: Festivals and Gold Prices
Some findings (details not shown here).
impact on. . .festival expectation volatility
Akshaya Tritiya * *
Chinese New Year *
Christmas * *
Diwali
Dussehra *
Easter *
Eid al-Adha *
Ramadan Eid * *
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 3 20/32
Project 4:
Population Dynamics With Leslie-Type Models
The classical Leslie model.
• Leslie model:
a discrete, age-structured model of population growth
• time-constant age-specific fertility and mortality rates
• population is closed to migration
• only females considered
• three 15-year intervals of age covering ages 0 to 45
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 4 21/32
Project 4:
Population Dynamics With Leslie-Type Models
The classical Leslie model.
• Equation of the Leslie model: n1,t
n2,t
n3,t
=
f1 f2 f3p1 0 0
0 p2 0
·
n1,t−1
n2,t−1
n3,t−1
~nt = M · ~nt−1
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 4 22/32
Project 4:
Population Dynamics With Leslie-Type Models
A Leslie-type model for a population with constant immigration.
n1,t
n2,t
n3,t
n∗1,t
n∗2,t
n∗3,t
R
=
f1 f2 f3 f∗1 f∗2 f∗3 0
p1 0 0 0 0 0 0
0 p2 0 0 0 0 0
0 0 0 0 0 0 r10 0 0 p∗1 0 0 r20 0 0 0 p∗2 0 r30 0 0 0 0 0 1
·
n1,t−1
n2,t−1
n3,t−1
n∗1,t−1
n∗2,t−1
n∗3,t−1
R
(Schmidbauer & Rosch, 1995)
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 4 23/32
Project 4:
Population Dynamics With Leslie-Type Models
A Leslie-type model with two populations.
nc1,t
nc2,t
nc3,t
n∗c1,t
n∗c2,t
n∗c3,t
nv1,t
nv2,t
nv3,t
n∗v1,t
n∗v2,t
n∗v3,t
=
♣ ♣ ♣ ♣ ♣ ♣ 0 0 0 0 0 0♠ 0 0 0 0 0 0 0 0 0 0 00 ♠ 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 ♣ ♣ ♣ ♣ ♣ ♣0 0 0 ♠ 0 0 ♠ 0 0 ♠ 0 00 0 0 0 ♠ 0 0 ♠ 0 0 ♠ 00 0 0 0 0 0 ♣ ♣ ♣ ♣ ♣ ♣0 0 0 0 0 0 ♠ 0 0 0 0 00 0 0 0 0 0 0 ♠ 0 0 0 0♣ ♣ ♣ ♣ ♣ ♣ 0 0 0 0 0 0♠ 0 0 ♠ 0 0 0 0 0 ♠ 0 00 ♠ 0 0 ♠ 0 0 0 0 0 ♠ 0
·
nc1,t−1
nc2,t−1
nc3,t−1
n∗c1,t−1
n∗c2,t−1
n∗c3,t−1
nv1,t−1
nv2,t−1
nv3,t−1
n∗v1,t−1
n∗v2,t−1
n∗v3,t−1
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 4 24/32
Project 4:
Population Dynamics With Leslie-Type Models
Turkey: Long-run growth and urbanization.
mv
mc
0.75 0.8
0.85
0.9
0.95
1
1.0
5
1.1
1.1
5
1.2
1.2
5
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
mv
mc
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 4 25/32
Project 5:
OPEC Announcements and Oil Price Volatility
Crude oil prices and OPEC announcements.
• Impact of OPEC announcements on crude oil prices?
• Impact on the distribution of daily returns, in particular:
– on the expectation of daily returns?– on the variance of daily returns?
• What can be said about expectation and volatility. . .
– right before an announcement will be made
(anticipation of the announcement),
– right after an announcement has been made
(aftereffect of the announcement)?
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 5 26/32
Project 5:
OPEC Announcements and Oil Price Volatility
The WTI price series and OPEC announcements.
050
100
150
US
D/b
arre
l
1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010
cut
maintain
increase
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 5 27/32
Project 5:
OPEC Announcements and Oil Price Volatility
The daily WTI return series and OPEC announcements.
−30
−20
−10
010
2030
daily
ret
urn
in p
erce
nt
1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010
cut
maintain
increase
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 5 28/32
Project 5:
OPEC Announcements and Oil Price Volatility
Regression: conditional expectation; GARCH: conditional variance.
rt = c+∑s≥1
asrt−s +∑i
bidit + εt, (4)
εt = νt ·√ht, (5)
ht = α0 + α1ε2t−1 + βht−1 +
∑i
γidit. (6)
• (rt): series of daily returns on WTI crude oil price
• (dit): (modified) dummy variables for announcements of kind i
• (νt): Gaussian white noise with var(νt) = 1
• bi, γi: parameters (impact of an announcement of kind i)
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 5 29/32
Project 5:
OPEC Announcements and Oil Price Volatility
Summary: The optimal model structure.
expectation volatility
cut● ● ● ● ● ●
● ●
● ● ● ● ● ● ● ●0.0
0.5
1.0
ta −− 6 ta −− 3 ta ta ++ 3 ta ++ 6 ta ++ 9
● ●●
●
●
● ●
● ● ● ● ● ● ● ● ●0.0
0.5
1.0
ta −− 6 ta −− 3 ta ta ++ 3 ta ++ 6 ta ++ 9
increase● ● ● ● ● ● ● ●
● ● ● ● ●
●
●●0.0
0.5
1.0
ta −− 6 ta −− 3 ta ta ++ 3 ta ++ 6 ta ++ 9● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●0.0
0.5
1.0
ta −− 6 ta −− 3 ta ta ++ 3 ta ++ 6 ta ++ 9
maintain● ● ● ● ● ● ●
●
● ● ● ● ● ● ● ●0.0
0.5
1.0
ta −− 6 ta −− 3 ta ta ++ 3 ta ++ 6 ta ++ 9● ● ●
●
●
●
● ● ● ● ● ● ● ● ● ●0.0
0.5
1.0
ta −− 6 ta −− 3 ta ta ++ 3 ta ++ 6 ta ++ 9
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 5 30/32
Project 5:
OPEC Announcements and Oil Price Volatility
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 5 31/32
Research.
The scientific journal list relevant for Turkey:
http://www.ulakbim.gov.tr/cabim/ubyt/dergilist.php
c© Harald Schmidbauer & Angi Rosch, 2012 From Our Recent Research: Project 5 32/32